Mapping mac address into elasticsearch index using jdbc river

Question

We have a type that contains mac address field. the data is brought using jdbc river

The cause is that when we run a term aggregation on the mac_address field the results looks like the field is broke into indexed keys:

Action:

GET index/type/_search?search_type=count
{
    "aggs" : { 
        "uniqe_macs" : { 
            "terms" : {
              "field" : "mac_address" 
            }
        }
    }
}

Result:

  "aggregations": {
     "uniqe_visitors": {
        "buckets": [
           {
              "key": "00",
              "doc_count": 1608759
           },
           {
              "key": "10",
              "doc_count": 674633
           },
           {
              "key": "18",
              "doc_count": 588591
           },
           {
              "key": "f0",
              "doc_count": 544897
           },
           {
              "key": "60",
              "doc_count": 538841
           },
           {
              "key": "40",
              "doc_count": 529085
           },
           {
              "key": "08",
              "doc_count": 523681
           },
           {
              "key": "d0",
              "doc_count": 515774
           },
           {
              "key": "54",
              "doc_count": 514771
           },
           {
              "key": "04",
              "doc_count": 509629
           }
        ]
     }
    }

What can be done to force elastic to map this field and not to break it into keys?

prayagupa · Accepted Answer · 2014-09-09T13:12:48.430

4

Can you try with following mapping, custom analyzer on es field mac_address.

Define analyzer

curl -XPUT http://localhost:9200/INDEX  -d '
{
    "settings" : {
        "analysis" : {
            "analyzer" : {
                "my_edge_ngram_analyzer" : {
                    "tokenizer" : "my_edge_ngram_tokenizer"
                }
            },
            "tokenizer" : {
                "my_edge_ngram_tokenizer" : {
                    "type" : "edgeNGram",
                    "min_gram" : "2",
                    "max_gram" : "17"
                }
            }
        }
    }
}'

Apply mapping

curl -XPUT http://localhost:9200/INDEX/TYPE/_mapping  -d '
{
    "TYPE": {
        "properties" {
            "mac_address": {
                "type": "string",
                "index_analyzer" : "my_edge_ngram_analyzer",
                "search_analyzer": "keyword"
            }
        }
    }
}'

edited Sep 09 '14 at 13:12

answered Sep 09 '14 at 13:07

prayagupa

30,204
14
155
192

Thanks for the answer, can I set an analyzer post indexing or do I need to fetch the data again? – Ben Diamant Sep 09 '14 at 13:18
Hi, You need to apply mapping before indexing just like sql schema. – prayagupa Sep 09 '14 at 13:51
if you remember correctly if you treat a string as a constant value and don’t want to analyze it then you don’t need an analyzer? – Ben Diamant Sep 09 '14 at 14:10
In such case your mapping will be "index" : "not_analyzed" instead of default analyzer as you said. – prayagupa Sep 09 '14 at 14:16

score 0 · Answer 2 · answered Jan 03 '15 at 11:50

For me it was easier to define a raw multifield for the mac_adress and set it to not_analyzed, as described here. Although it did not work for the old data, no need to alter the index with a new analyzer.

curl -XPUT http://localhost:9200/INDEX/TYPE/_mapping -d'

{
    "TYPE" : {
        "properties" : {
            "mac_address" : {
                "type" : "string",
                "fields":{
                    "raw" : {
                      "type": "string",
                      "index": "not_analyzed"
                    }
                  }
            }
        }
    }
}'

Then for the aggregation you just need to use the field mac_address.raw:

curl -XPOST http://localhost:9200/INDEX/TYPE/_search?search_type=count -d'

{
    "aggs" : { 
        "unique_macs" : { 
            "terms" : {
              "field" : "mac_address.raw" 
            }
        }
    }
}'

Mapping mac address into elasticsearch index using jdbc river

2 Answers2